Redis Open Source 8.8 release notes

===================================

--------------------------------------------------------------------------------

Upgrade urgency levels:

LOW: No need to upgrade unless there are new features you want to use.

MODERATE: Program an upgrade of the server, but it's not urgent.

HIGH: There is a critical bug that may affect a subset of users. Upgrade!

CRITICAL: There is a critical bug affecting MOST USERS. Upgrade ASAP.

SECURITY: There are security fixes in the release.

--------------------------------------------------------------------------------

The release notes contain PRs from multiple repositories:

#n - Redis (https://github.com/redis/redis)

#Qn = Query Engine (https://github.com/RediSearch/RediSearch)

#Jn = JSON (https://github.com/RedisJSON/RedisJSON)

#Tn = Time Series (https://github.com/RedisTimeSeries/RedisTimeSeries)

#Pn = Probabilistic (https://github.com/RedisBloom/RedisBloom)

================================================================================

Redis 8.8.3 Released Thu 17 Sep 2026 18:00:00 IST

================================================================================

Update urgency: `SECURITY`: There are security fixes in the release.

### Security fixes

- #15673 Commands queued in a transaction could still access keys whose ACL permissions were revoked before the transaction was executed

- #15722 The cluster bus protocol has no authentication of its own unless `tls-cluster` is enabled, so any host able to reach a node's bus port could join the cluster and threaten it. A cluster node now warns at startup when its bus port is left unauthenticated, and the new `cluster-bus-port-protected-mode` option (default `no`) makes refusing to run in that state an explicit choice: set it to `yes` and the node starts only when `tls-cluster` authenticates the bus

- TimeSeries: Prevented Redis from crashing when adding samples to a compressed Time Series key restored from a malformed RDB payload

- Vector Sets: Deeply nested JSON used in Vector Set queries could cause the server to crash

================================================================================

Redis 8.8.2 Released Mon 17 Aug 2026 10:00:00 IST

================================================================================

Update urgency: `SECURITY`: There are security fixes in the release.

### Security fixes

- (CVE-2026-62356) Miscalculated buffer size in `CMSketch` RDB loading may lead to heap OOB write

- Out-of-bounds access in TopK heap cleanup path (MOD-15410)

- Use-after-free in the TLS pending-data list when a command closes another pending connection

- #15478 ACL key permission bypass in `SORT`, `GEORADIUS`/`GEORADIUSBYMEMBER` and `XREAD`/`XREADGROUP`: the keys validated by ACL could differ from the keys the command actually accesses

- A malicious RDB payload with an out-of-range `SLOT_INFO` slot id causes memory corruption during RDB loading, which may lead to Remote Code Execution

- Vector Sets: missing node level validation when loading a vector set from RDB may lead to out-of-bounds access

- Vector Sets: use-after-free when `VREM` mutates the HNSW graph while background `VSIM` threads are still running

- Vector Sets: a negative `hnsw_search()` return was treated as a huge unsigned count, reading past the end of the result arrays

- TLS client certificate authentication bypass: a Common Name containing an embedded NUL byte was truncated, allowing a client to authenticate as another (possibly privileged) ACL user

- #15594 Use-after-free in the blocked-client list when reprocessing a command evicts another client blocked on the same key

================================================================================

Redis 8.8.1 Released Thu 23 Jul 2026 22:00:00 IST

================================================================================

SECURITY: There is a security fix in the release.

### Security fixes

- #P1044 Crafted RESTORE payloads in RedisBloom and TDigest may trigger out-of-bounds writes, potentially leading to remote code execution

=================================================================

Redis 8.8 GA (v8.8.0) Released Mon 25 May 2026 12:00:00 IST

=================================================================

This is the General Availability release of Redis 8.8 in Redis Open Source.

### Major changes compared to 8.6

- New data structure: Array (@antirez)

- Subkey notification for hash fields - field-level notifications

- `INCREX`: a window counter rate limiter combining `INCR`, `INCRBY`, `INCRBYFLOAT`, bounds, and expiration (@raffertyyu + Redis team)

- `XNACK`: a new streams command - allow consumers to explicitly release pending messages

- `ZUNION`, `ZINTER`, `ZUNIONSTORE`, `ZINTERSTORE`: new `COUNT` aggregator

- `JSON.SET`: new `FPHA` argument to specify the FP type for homogeneous FP arrays

- `TS.RANGE`, `TS.REVRANGE`, `TS.MRANGE`, `TS.MREVRANGE`: multiple aggregators in a single command

- `FT.HYBRID` `KNN` clause: new argument to request fewer candidates per shard

- `FT.PROFILE` `HYBRID`: profiling support for `FT.HYBRID`

- Performance improvements

### Binary distributions

- Alpine and Debian Docker images - https://hub.docker.com/_/redis

- Install using snap - see https://github.com/redis/redis-snap

- Install using brew - see https://github.com/redis/homebrew-redis

- Install using RPM - see https://github.com/redis/redis-rpm

- Install using Debian APT - see https://github.com/redis/redis-debian

### Operating systems we test Redis 8.8 on

- Ubuntu 22.04 (Jammy Jellyfish), 24.04 (Noble Numbat), 26.04 (Resolute Raccoon)

- Rocky Linux 8.10, 9.7, 10.1

- AlmaLinux 8.10, 9.7, 10.1

- Debian 12.13 (Bookworm), Debian 13.4 (Trixie)

- Alpine 3.23

- macOS 14.8.4 (Sonoma), 15.7.4 (Sequoia), 26.3 (Tahoe) - for both Intel and ARM

### Bug fixes (compared to 8.8-RC1)

- #15237 `INCREX` syntax update

- #15005 Memory tracking can be enabled at runtime in non-clustered mode

- #T1930 Cluster topology changes during a multi-shard command are not handled (MOD-14439)

- #P1007 Memory leak on RDB load (MOD-15418)

=================================================================

Redis 8.8-RC1 (v8.7.240) Released Thu 14 May 2026 18:00:00 IST

=================================================================

This is the first Release Candidate of Redis 8.8 in Redis Open Source.

Release Candidates are feature-complete pre-releases. Pre-releases are not suitable for production use.

### Headlines:

Redis 8.8 introduces new features and performance improvements.

### Operating systems we test Redis 8.8 on

- Ubuntu 22.04 (Jammy Jellyfish), 24.04 (Noble Numbat), 26.04 (Resolute Raccoon)

- Rocky Linux 8.10, 9.7, 10.1

- AlmaLinux 8.10, 9.7, 10.1

- Debian 12.13 (Bookworm), Debian 13.4 (Trixie)

- Alpine 3.23

- macOS 14.8.4 (Sonoma), 15.7.4 (Sequoia), 26.3 (Tahoe) - for both Intel and ARM

### Security fixes (compared to 8.8-M03)

- (CVE-2026-23479) Use-After-Free in unblock client flow may lead to Remote Code Execution.

- (CVE-2026-25243) Invalid memory access in `RESTORE` may lead to Remote Code Execution

- (CVE-2026-23631) Lua Use-After-Free may lead to remote code execution

- (CVE-2026-25588) Invalid memory access in `RESTORE` may lead to Remote Code Execution (Time Series)

- (CVE-2026-25589) Invalid memory access in `RESTORE` may lead to Remote Code Execution (Probabilistic)

### New Features (compared to 8.8-M03)

- #15162 New data structure: Array (@antirez)

- #15045 `INCREX`: a window counter rate limiter combining `INCR`,`INCRBY`,`INCRBYFLOAT`, bounds, and expiration (@raffertyyu + Redis team)

- In group sorting new reducer, allowing unwind grouped documents (after `GROUPBY`) and sort them

### Removed Features (compared to 8.8-M03)

- #15191 Remove GCRA rate limiter

### Bug fixes (compared to 8.8-M03)

- `SUBSCRIBE`, `PSUBSCRIBE`, `SSUBSCRIBE`: crash on OOM (RED-167788)

- `CONFIG SET`: some settings allow invalid characters (RED-167787)

- `SCRIPT DEBUG`: potential crash on scripts (RED-175507)

- `VADD`: crash or buffer overflow on large `REDUCE` value (RED-170921)

- `VSET`: crash on huge allocations (MOD-12678)

- #15188 `cluster-announce-ip` rejecting hostnames (regression)

- #15095 Double free when loading streams with duplicate consumer PEL entries

- #15124 Issues processing corrupt Streams RDB data

- #15111 `fast_float_strtod` rounding mismatch

- #15190 `vecClear` reset the logical size without releasing element ownership

- #15163 `MULTI` queue memory incorrect memory accounting

- #15094 Cluster crash when `CLIENT KILL` unsubscribes `SSUBSCRIBE` client inside `EXEC`

- #15151 Listpack backlength encoding thresholds off-by-one

- #15115 Under-copy in the Lua debugger

- #14970 Sentinel config injection via `SENTINEL SET`

- #14934 Client output buffer memory tracking not accounting for copy-avoided bulk string references

- #Q9182 `FT.PROFILE HYBRID` returns an empty reply (MOD-14778)

- #Q9079 `FT.SPELLCHECK` treats `PARAMS` placeholders as literal terms instead of resolving them (MOD-10596)

- #Q9047 `FT.PROFILE` output is inconsistent when a profiled value is missing (MOD-10560)

- #Q9078 `FT.CREATE` now rejects schema definitions with invalid option combinations at creation time (MOD-14655)

- #Q9012 `PERSIST` and `HPERSIST` notifications are not reflected in index expiration tracking (MOD-14800)

- #Q9066 Race condition in `FT.HYBRID` causes intermittent failures under concurrent hybrid query load (MOD-14732)

- #Q9163 Crash on `FT.SEARCH` when topology validation fails (for example, some nodes unreachable) (MOD-14475)

- #Q9031, #Q9473 Coordinator deadlock under mixed `FT.SEARCH` and `FT.AGGREGATE` load (MOD-14268)

- #Q9028 Memory leak when `FT.DROPINDEX` runs concurrently with in-flight hybrid queries (MOD-14135)

- #Q9310, #Q9350 `FT.CURSOR READ` timeout and `ON_TIMEOUT FAIL` not enforced on coordinator and shard (MOD-14284, MOD-14998)

- #Q9425 Cursors not cleaned up after `MAXIDLE`, causing resource exhaustion (MOD-6430)

- #Q9234, #Q9404 Coordinator `RETURN_STRICT` returns wrong data on partial results, including `SORTBY` pipeline (MOD-13617)

- #Q9382 `MAXPREFIXEXPANSION` warnings not propagated to clients in cluster mode (MOD-13804)

- #Q9218 Search commands fail when no worker thread is available instead of falling back to main thread (MOD-14921)

- #Q9448 RDB load missing validation of `FT.CREATE` arguments, allowing corrupt index state on load (MOD-13118)

- #Q9377 Use-after-move in `Indexer_Process` causes crash during indexing (MOD-14980)

- #Q9408 Deadlock between background query and main-thread writer (MOD-15364)

- #Q9114 `FT.PROFILE` prints output using wrong iterator type (MOD-14678)

- #Q9421 Confusing error returned when `DEBUG_PARAMS_COUNT` is zero (MOD-15118)

- #Q9045 Stack-smashing error in coordinator code path (MOD-14649)

- #J1554 Trailing chars are ignored (MOD-7266); Fixes #J976

- #J1543 Wrong mutation ordering for array commands with recursive paths (MOD-6722)

- #J1542 JSONPath evaluation issues (MOD-14664); Fixes #J968 (MOD-7264), #J962 (MOD-7272), #J963 (MOD-7270), #J1089 (MOD-7268)

- #T2003 Potential crash on disconnections and TLS failures (MOD-14850)

- #T2013 `count`, `countNaN`, `countAll` reducers return NaN when all values are NaN (MOD-14420)

### Performance and resource utilization improvements (compared to 8.8-M03)

- #15049 Hyperloglog: 4 independent accumulators that are merged at the end

- #15133 Batched prefetch for `MGET` and `MSET`

- #14988 Batched prefetch for `HGETALL` on hashtable-encoded hashes

- #15071 Pass size hint to jemalloc for faster deallocation

- #15096 Reduces allocator and accounting overhead by adding compile-time jemalloc tuning

- #Q9197 Vector index hot path (HNSW and brute-force) devirtualized, reducing per-query latency (MOD-14916)

- #Q9262, #Q9476 Inline LSE atomics enabled on AArch64, improving atomic operation throughput on ARM64 (MOD-14916, MOD-15419)

- #Q9293 Expiration handling overhead reduced when many keys expire simultaneously (MOD-14916)

- #Q9017 LTO (link-time optimization) enabled for x86_64 release builds (MOD-14700)

- #Q8765 Shard-level timeout adjusted to coordinator dispatch time for more accurate accounting (MOD-13189)

- #Q8790, #Q8900, #Q8827, #Q8971, #Q8966, #Q8762, #Q8678, #Q8915, #Q8653, #Q9085, #Q8751, #Q8692, #Q9224 Iterators ported to Rust, reducing FFI overhead

- #Q9500 `numRecords` no longer updated for vector fields, removing unnecessary write overhead on ingest (MOD-15487)

- VecSim SVS thread pool integrated with the worker pool for better thread utilization (MOD-9881)

### Configuration parameters

- #15182 Slowlog entry truncation limits:

- `slowlog-entry-max-argc`: maximum number of command arguments kept in a slowlog entry

- `slowlog-entry-max-string-len`: maximum length of a command argument in a slowlog entry

- #Q8876, #Q8960 Default maximum worker threads value updated; `MAX_WORKER_THREADS` is now a string config (MOD-14486, MOD-14763)

### Metrics (compared to 8.8-M03)

- #Q8210, #Q8231 `FT.PROFILE`: added queue time tracking (MOD-13602)

### CLI tools

- #15150 Memory leak on malformed legacy help entry in redis-cli

=================================================================

Redis 8.8-M03 (v8.7.226) Released Tue 28 Apr 2026 16:00:00 IST

=================================================================

This is the third Milestone of Redis 8.8 in Redis Open Source.

Milestones are non-feature-complete pre-releases. Pre-releases are not suitable for production use.

### New Features (compared to 8.8-M02)

- #14958 Subkey notification for hash fields - field-level notifications

- #Q8227 `FT.HYBRID` `KNN` clause: new argument to request fewer candidates per shard

- #Q8060 `FT.PROFILE` `HYBRID`: profiling support for `FT.HYBRID`

### Bug fixes (compared to 8.8-M02)

- #15034, #15081 Issues processing corrupt RDB data

- #15059 Use-after-free

- #15073 `CLIENT TRACKING`: self-overlap returning non-zero loop index

- #14982 `SCAN` commands: integer overflow in `COUNT` parameter

- #14956 Crash on `HSETEX` when a field appears more than once and an expiry is specified

- #15015 Change log level for unknown extension types from `LL_WARNING` to `LL_VERBOSE`

- #14995 Unnecessary `-ERR` and `\r\n`

- #Q8708 Crash when many keys receive expirations under heavy TTL activity

- #Q8774 Coordinator deadlock under mixed `FT.SEARCH` and `FT.AGGREGATE` load

- #Q8415 Crash on `FT.SEARCH` when topology validation fails (for example, some nodes unreachable)

- #Q8322 Crash when indexing negative zero (-0.0)

- #Q8843 HNSW vector index memory growth under high-churn workloads until shard restart

- #Q8396 `FILTER` returns inconsistent results with multiple indexes sharing field aliases

- #Q8205 `FT.HYBRID` `VSIM RANGE` + `FILTER` incorrectly returns zero results

- #Q8817 Instability and crashes in long-running search cursors during concurrent index updates

- #Q8388 `FT.SEARCH` fails with “Query requires unavailable slots” after shard restart or failover

- #Q8548 `FILTER` behavior depends on property order in the expression

- #Q8320 Index `FILTER` applied inconsistently when documents are missing filtered fields

- #Q8752 Missing blocked-client FAIL timeout mechanism for coordinator-level `FT.AGGREGATE`

- #Q8657 Missing shard-level FAIL timeout handling for `FT.HYBRID` queries

- #Q8420 Missing coordinator-level FAIL timeout handling for `FT.HYBRID` queries

- #Q8335 Legacy shard-level FAIL handling for `FT.SEARCH` / `FT.AGGREGATE`

- #Q8191 `FT.SEARCH` coordinator lacks strict FAIL timeout enforcement

### Performance and resource utilization improvements (compared to 8.8-M02)

- #15114 Optimize SET key value GET

- #15065, #15118 Scan commands key collection: replace list with append-only pointer vector

- #15061 Widen `fast_float_strtod` fast path to 17-19 digit mantissas

- #Q8378 Optimize filter expression evaluation: skip indexes not matching the document type (MOD-14064)

### Metrics (compared to 8.8-M02)

- #Q8246 ‘frontend_buffer_size’, ‘HNSW_main_thread_insertion’: metrics for tiered vector indexes (MOD-13819)

- #Q8210 `FT.PROFILE`: added queue time tracking (MOD-13602)

- #Q8283 `INFO`: Skip metrics when there are no indices (MOD-13903)

- #Q7417 Add unique error message ids for improved debugging and troubleshooting (MOD-11806)

### Configuration parameters

- #Q8876 ‘search-workers’: change default to 16 (MOD-14486)

- #Q8352 `BG_INDEX_SLEEP_DURATION_US`: sleep duration during background indexing (MOD-13994)

=================================================================

Redis 8.8-M02 (v8.7.225) Released Thu 16 Apr 2026 16:00:00 IST

=================================================================

This is the second Milestone of Redis 8.8 in Redis Open Source.

Milestones are non-feature-complete pre-releases. Pre-releases are not suitable for production use.

### Headlines:

Redis 8.8 introduces new features and performance improvements.

8.8-M02 is available as a Docker image and can be downloaded from [Docker Hub](https://hub.docker.com/_/redis). Additional distributions will be introduced in upcoming pre-releases.

### Operating systems we test Redis 8.8 on

- Ubuntu 22.04 (Jammy Jellyfish), 24.04 (Noble Numbat), 26.04 (Resolute Raccoon)

- Rocky Linux 8.10, 9.7, 10.1

- AlmaLinux 8.10, 9.7, 10.1

- Debian 12.13 (Bookworm), Debian 13.4 (Trixie)

- Alpine 3.23

- macOS 14.8.4 (Sonoma), 15.7.4 (Sequoia), 26.3 (Tahoe) - for both Intel and ARM

### New Features (compared to 8.6)

- #14826, #14905 GCRA (generic cell rate algorithm) rate limiter (based on the redis-cell module by @brandur - thank you!)

- #14797 `XNACK`: a new streams command that allows consumers to explicitly release pending messages

- #J1509 `JSON.SET`: new `FPHA` argument to specify the FP type for homogeneous FP arrays (MOD-13577)

- #14892 `ZUNION`, `ZINTER`, `ZUNIONSTORE`, `ZINTERSTORE`: new `COUNT` aggregator

- #T1916 `TS.RANGE`, `TS.REVRANGE`, `TS.MRANGE`, `TS.MREVRANGE`: multiple aggregators in a single command (MOD-9162)

### Bug fixes (compared to 8.6.2)

- #15037 `XINFO STREAM`: wrong value in the per-slot memory tracking

- #12000 Incorrect shrinking of querybuf when client is reading a big argv

- #15021 `HEXPIRE`: overflow on fields count

- #14963 `XREADGROUP`: consumer replication inconsistency

- #15011 Double-free in rdbLoadObject

- #14667, #14886 Potential TCP stalls/deadlocks

- #14942 Fix `COMMAND GETKEYS for PFMERGE` with no source keys

- #14888 Hardens safety check in `lpSafeToAdd`

- #14748 Ensure sensitive user data is not exposed in logs when `hide_user_data_from_log` is enabled

- #14877 `INFO KEYSIZES` and INFO ALLOCSIZES` fixes

- #14878 listpack memory leak in zipmap-to-hash conversion on error path

- #14955 Streams: IDMP-related bug

- #14974, #14932, #14866 Memory leaks

- #14790 Missing initialization

- #14789 DB hash tables not expanding during RDB load on standalone-mode

- #14785 Add `HOTKEYS HELP` subcommand (Each command having subcommands needs a `HELP` subcommand)

- #14784 `RESTORE` and `SET`: `expiredkeys` counter updates incorrectly on keys with past expiration time

- #14779 Accurate memory accounting for RedisModuleDict (instead of estimation)

- #14774 Missing cleanup code in `getKeysFreeResult()` on cross-slot error path

- #14771 Potential unsigned integer underflow in `used_memory_dataset calculation`

- #14780 Fixed name of `HOTKEYS` results field `sampled-command-selected-slots-us` to `sampled-commands-selected-slots-us`

- #14756 `HOTKEYS` does not track each command in a `MULTI`/`EXEC` block

- #14749 Fixed `HOTKEYS GET` RESP3 reply fixed to map instead of flat array

- #J1520 Numeric operations on homogeneous arrays: promote to next type if results overflow (MOD-14427)

- #J1519 `JSON.NUMINCRBY`, `JSON.NUMMULTBY` don’t error on numeric overflow (MOD-14191)

- #J1487 Depth limit (128) was not fully enforced (MOD-4107)

- #T1910 `TS.INCRBY`, `TS.DECRBY` create key before validating args (MOD-8167)

- #T1908 Potential crash on `TS.RANGE` with aggregation and `EMPTY`

- #T1896 Potential memory leak (MOD-13438)

- #T1913 ACL rules are not fully enforced for multi-key commands on a cluster setup (MOD-14124)

### Performance and resource utilization improvements (compared to 8.6.2)

- #15003 Reduce overhead in command propagation

- #14979 Dismiss dict bucket arrays in fork child to reduce CoW

- #14661 Replace fast_float C++ library with pure C implementation

- #14928 Disable memory tracking in child processes

- #14885 Optimize rax (radix tree) insert and lookup for sequential key patterns

- #14335 Handle primary/replica clients in I/O threads

- #14754 Optimize prefetching commands

- #14770 Optimize `BITOP` operations with AVX512

- #14862 Streams: Filter expired IDMP entries during RDB save and load

- #14750 `SFLUSH` can now flush slots partially

- #14887 `SFLUSH` skips slot trimming when the requested slot ranges exactly match the node’s local slot coverage

- #14851 Skip RDB checksum computation/validation during diskless full synchronization to reduce CPU overhead

- #14783 Pause dict auto-resize during multi-field deletion

- #1884 `TS.RANGE`, `TS.REVRANGE`, `TS.MRANGE`, `TS.MREVRANGE`: improve performance on a cluster setup (RED-184104)

- #T1866 Register to `type_changed` KSN instead of `set` (MOD-12919)

### Modules API

- #14445 Allow modules to associate metadata with keys

- `RedisModule_CreateKeyMetaClass` - define a new key-metadata class

- `RedisModule_ReleaseKeyMetaClass` - release a key-metadata class

- `RedisModule_SetModuleMetadata` - attach or update metadata to a key

- `RedisModule_GetKeyMeta` - get key metadata

### Metrics

- #14896 `INFO STATS` - global stats for slowlog metrics:

- `slowlog_commands_count` - commands written to slowlog

- `slowlog_commands_time_ms_sum` - sum of execution times of commands from the slowlog

- `slowlog_commands_time_ms_max` - maximum execution time of a command from the slowlog

- #14896 `INFO COMMANDSTATS` - per-command stats for slowlog metrics:

- `slowlog_count` - number of times the command was written in the slowlog

- `slowlog_time_ms_sum` - sum of execution time of the command (only from the slowlog)

- `slowlog_time_ms_max` - maximum execution time of the command (only from the slowlog)

-14841 `INFO STATS` (global, all clients aggregated):

- `total_client_processing_events`: attempts to process client input buffers; does not guarantee any command was actually parsed

- `eventloop_cycles_with_clients_processing`: event loop cycles where client input buffers were processed

- `commands_per_parse_batch_sum`: cumulative number of commands parsed across all parsing batches for all clients

- `commands_per_parse_batch_cnt`: number of parsing batches across all clients. A batch is counted each time at least one command is parsed from a client's query buffer

- `commands_per_parse_batch_avg`: average commands parsed per batch (sum/cnt). Approximates pipelining depth

-#14841 `CLIENT INFO` and `CLIENT LIST` (per-client):

- `read-events`: number of read events for this client

- `parse-batch-cmd-sum`: cumulative number of commands parsed across all parsing batches for this client

- `parse-batch-cnt`: total number of parsing batches for this client. Divide `parse-batch-cmd-sum` by this value to get the client’s average commands per batch

### CLI tools

- #14371 Divide-by-zero in redis-benchmark and redis-cli

- #14863 Memory leak in redis-cli

- #14703 redis-cli --keystats: fix percentile calculation